You are on page 1of 4

2012 4th International Conference on Intelligent and Advanced Systems (ICIAS2012)

Implementation of Cascade Control for Wheeled


Mobile Robot Straight Path Navigation
K.H. Ng, C.F. Yeong, E.L.M. Su A.R. Husain
Faculty of Electrical Engineering, Centre for Student Innovation,
Universiti Teknologi Malaysia Office of Undergraduate Studies,
81310 Skudai, Johor, Malaysia Universiti Teknologi Malaysia
81310 Skudai, Johor, Malaysia

Abstract - A cascade controller was proposed for mobile robot There are cases where nonlinear controllers were used in
navigation because of the simplicity of control algorithm. For a trajectory tracking like [3] where the dynamic tracking
robot that requires to only complete straight path navigation, the
cascade controller is an attractive solution. Compared to typical stabilization was analysed using Lyapunov function while [4]
mobile robot navigation algorithms, this cascade control does not used sliding mode controller for the tracking of nonholonomic
need any path tracking control and only uses simple PID for mobile robot.
navigation. For each mobile robot wheels, a position PID All given examples above involve a planned desired path
controller is cascaded to a velocity PID controller. An additional and a control law so that the mobile robot will navigate
PID controller is added in between both wheels to make sure the
mobile robot moves in a straight path. An experiment was done according to the planned path. All these algorithms are either
on a differential drive robot and results were presented. simulations or implemented in a computer.
In the ABU ROBOCON [5], mobile robots are required to
I. INTRODUCTION navigate throughout the field. Universiti Teknologi Malaysia
Navigation of wheeled mobile robot usually includes path has participated in this competition and developed automatic
planning and path tracking. For path planning, the acceleration mobile robots for the game. The automatic robot has the
and deceleration profile is considered and the trajectory is ability to move in straight path on the game field as shown in
generated in a Cartesian coordinate. Fig. 1.
Thereafter, control laws are developed to track this Due to the implementation of mobile robot programming in
trajectory. There are several examples where linear controllers microcontroller, the above mentioned methods are impractical
were used, for example in [1], PID controller was used to track due to computation time and memory. Therefore, a cascade
the heading of a synchronous robot based on desired controller was proposed where only the desired position and
trajectory. While in [2], PI controller with feedforward velocity are fed into the controller and the mobile robot will
compensator was used to track the desired trajectory. move in a straight path. There is no path tracking control
needed for this method and the tracking only involves the
cascade controller which makes it quite easy to be
implemented in a microcontroller.
The differential drive wheeled mobile robot kinematic
model is used to analyse the robot posture in world frame. In
other words, it is used to show where the mobile robot has
travelled. Experiments were conducted to validate the
proposed cascaded controller. Conclusion and
recommendation for future works are presented at the end of
the paper.
II. WHEELED MOBILE ROBOT KINEMATIC MODEL
Mobile robots can be classified to many types based on the
degree of steerability and degree of mobility [6]. The type
(2,0) was chosen because of the simple kinematic model used
and the simplicity in robot fabrication.

Figure 1. Part of game field for the contest

978-1-4577-1967-7/12/$26.00 ©2011 IEEE

[ 503 ]
2012 4th International Conference on Intelligent and Advanced Systems (ICIAS2012)

Figure 3. Cascaded controller scheme

Figure 2. Mobile robot posture used in this experiment

A differential drive robot is classified under type (2,0). The


model for this differential drive robot posture is used in many
mobile robot [2], [4], [7]. As shown in Fig. 2, the mobile robot
consists of two wheeled placed in parallel and from the
encoder readings, the forward kinematics can be used to obtain
the position of the robot in the Cartesian coordinate frame.
As the algorithm in a microcontroller uses the numerical
computation, the forward kinematics will be implemented
using the Euler equations based on the formula in [7]. The
equations are:

r (ω r + ω l ) r (ω r + ω l )
v= ,ω = (1)
2 d
Figure 4. Full system for the mobile robot navigation
where r is the radius of the wheel and d is the distance of
both wheels, ω r and ωl are both right and left wheel velocity. By doing so, the mobile robot will move to a desired
position in a straight path with a desired velocity. This
xk +1 = xk + vk Ts cos(θ k ) (2) condition is possible if the desired position is relatively larger
y k +1 = yk + vk Ts sin(θ k ) (3) than then desired velocity.
An additional PID controller is added into the system to
θ k +1 = θ k + ωk Ts (4)
compensate for position difference. The controller is inserted
in between the two controllers like shown in Fig. 4. This type
where Ts is the fixed sampling time. of compensation is coined in [10] where only an integrator
This kinematic model is used only as an analysis method for was used. But for better control, a full PID is used for this
this experiment because the model does not influence the purpose.
controller output to the actuators.
IV. EXPERIMENTAL RESULTS
III. CASCADE CONTROLLER
For the experiment, a robot was fabricated from aluminum
Cascaded PID controller can be found in many literatures in bars, direct current motors and quadrature encoders. The total
water level control where the inner loop represents the flow weight of the robot is 8.6 kg with length of 0.5m and wheels
control and while the outer loop represents the level control. distance of 0.7m. The robot is shown in Fig. 5. The robot will
Based on [8], [9] cascade control has the advantage of faster move on vinyl flooring as required by the ABU ROBOCON
response and better disturbance rejection. contest.
The proposed controller for straight path navigation consists The cascade controller is implemented in a microcontroller
of a pair of controller for each wheel. Each pair will have a using the UTM team’s motor control board where the cascade
PID controller for position and a PID controller for velocity. control, encoder interface and motor actuation are
The position controller output will be saturated to a desired programmed in a dsPIC33FJ128MC802 microcontroller. The
velocity before being fed into the velocity controller. The feedback was obtained from quadrature encoders attached to
proposed controlled is shown in Fig. 3. the ground and the actuators are 41 Watt direct current motors.

[ 504 ]
2012 4th International Conference on Intelligent and Advanced Systems (ICIAS2012)

Firstly, the inner loop of the cascade controller, which is the


velocity controller, was tuned.. The tuning method applied is
online trial and error method. The desired speed will be fed
with 0.5 m/s for two seconds thhen, 0.1 m/s for 4 seconds. The
response graph is shown in Fig. 6.
After that, the outer loop waas tuned to perform the straight
path navigation for the mobilee robot. Again, online trial and
error tuning method was usedd. The desired position is two
meters and the desired speed saaturation was set to 1.5 m/s.
The response graph is show wn in Fig. 7. The mobile robot
exhibits acceleration and decelleration profile which is similar
to typical mobile robot navigation with trapezoidal velocity
profile. This can be seen from m the velocity graph. And the
desired velocity is closed to thee desired velocity of 1.5 m/s.
Figure 5. Actual robot used for the expeeriment.
To analyze the position of o mobile robot in Cartesian
coordinate, the kinematic moddel was used and the position is
shown in Fig. 8.
Thereafter, the position diffference controller structure was
tested. Desired position of twoo meters and desired velocity of
0.5 m/s was fed into the conttroller. While the mobile robot
was navigating, external forces were exerted to the robot for
three times.

Figure 6. Mobile robot velocity responsee graph

Figure 8. Robot posturee in Cartesian coordinate

Figure 7. Mobile robot position responsee graph


Figure 9. Position respponse with disturbance

[ 505 ]
2012 4th International Conference on Intelligent and Advanced Systems (ICIAS2012)

The position response graph was shown inn Fig. 9. From the V. CONCLUSION AND
D FUTURE WORKS
robot position plot, both wheels were com mpensated so they
A cascade controller was prroposed and implemented for a
move in the same position. As shown in thhe velocity graph,
mobile robot to move in a straaight path. This algorithm only
there are oscillations but the average still remains
r about 0.5
needs desired position and veloocity for straight path movement
m/s. In other words, the position difference controller
c is trying
to be executed, as compared too navigation that involved path
to compensate for the difference by adjusting the velocity of
tracking control. The computaation for wheeled mobile robot
both motors.
navigation can be reduced siggnificantly. As for the tracking,
Again, to view the robot posture in the Carrtesian coordinate,
only PID controller was used annd the implementation is simple
the kinematic model was used to calculate the t robot position.
relatively to various tracking allgorithm [1–4].
This is shown in Fig. 10. Even there are sliight deviations out
For future works, the mobilee robot model will be identified
of the straight path, it was compensated backk.
using system identification toools. Proper PID tuning method
The weakness of this controller is it will not be able to reject
should be used like the Zieglerr-Nichols tuning method [12] to
large disturbance in the path, e.g. a distuurbance caused by
optimize the PID gains. Sinnce the controlling does not
robot-to-robot collision. This disturbance will
w make the robot
involved tracking, the Ziegler--Nichols tuning method should
deviate too much from the straight path and the robot will not
be sufficient.
know of it, ending up moving in a path parallel with the
original intended path, offset by a certain disstance. This case is REFER
RENCES
depicted in Fig. 11. This shows the weaknness in open loop
navigation scheme compared to closed loop navigation for [1] J. E. Normey-Rico, I. Alcalá, J. J Gómez-Ortega, and E. F. Camacho,
example in [11]. “Mobile robot path tracking ussing a robust PID controller,” Control
Engineering Practice, vol. 9, no. 11,
1 pp. 1209-1214, Nov. 2001.
[2] G. Mester, “Motion control of wheeled
w mobile robots,” in 4th Serbian-
Hungarian Joint Symposium onn Intelligent Systems, SISY, 2006, pp.
119–130.
[3] A. Gholipour and M. Yazdanppanah, “Dynamic tracking control of
nonholonomic mobile robot with w model reference adaptation for
uncertain parameters,” in Europpean Control Conference. Cambridge,
2003.
[4] Jung-Min Yang and Jong-Hwan Kim, “Sliding mode motion control of
nonholonomic mobile robots,” IE EEE Control Systems, vol. 19, no. 2, pp.
15-23, Apr. 1999.
[5] “ABU Robocon 20011.” [Online]. Available:
http://www.aburobocon2011.com m/. [Accessed: 30-Nov-2011].
[6] G. Campion and W. Chung, “Whheeled robots,” in Springer Handbook of
Robotics, B. Siciliano and O. Khhatib, Eds. Berlin, Heidelberg: Springer
Berlin Heidelberg, 2008, pp. 391--410.
[7] A. Luca, G. Oriolo, and M. Veendittelli, “Control of wheeled mobile
robots: An experimental overview w,” in Ramsete, vol. 270, S. Nicosia, B.
Figure 10. Robot posture in movement with disturbances
d Siciliano, A. Bicchi, and P. Vaaligi, Eds. Berlin, Heidelberg: Springer
Berlin Heidelberg, 2001, pp. 181--226.
[8] K. J. Åström and T. Hägglund, PID P controllers. International Society for
Measurement and Control, 1995.
[9] A. Visioli, Practical PID control. Springer, 2006.
[10] T. Bräunl, Embedded robotics: mobile robot design and applications
with embedded systems. Springerr, 2008.
[11] P. Morin and C. Samson, “Motioon control of wheeled mobile robots,” in
Springer Handbook of Robotics, B. Siciliano and O. Khatib, Eds. Berlin,
Heidelberg: Springer Berlin Heiddelberg, 2008, pp. 799-826.
[12] J. Ziegler and N. Nichols, “Optimmum settings for automatic controllers,”
Journal of dynamic systems, meaasurement, and control, vol. 115, p. 220,
1993.

Figure 11. Example of deviation from intended pathh due to disturbance

[ 506 ]

You might also like